Vector database sharding
Multimodal search
JaguarDB quantization
JaguarDB Vector API
Best Vector databases
JaguarDB in Docker
Setup JaguarDB with tar package
Setup JaguarDB on multiple nodes
Vector index sharing
How zeromove works
Video introduction
|
|
Example: Search on Time-Axis
Vector similarity search along a time-axis is a data retrieval technique that allows you to find data points or records that are similar to a given query vector while considering the temporal aspect. This approach enables you to retrieve data entries that share similar characteristics or patterns with the query vector within a specific time window or interesting time period. You specify an interesting time period or time window within which you want to search for similar data. For example, you may want to find data points that are similar to the query vector within the past week or month.
The following pseudo-code shows how to find similar images within a time window:
create store health (v vector(512, 'euclidean_fraction_float'), xray file, year date)
embeddings = getImageBeddings("xrayexam101.jpg")
insert into health values ('embeddings', '2015-09-18')
embeddings = getImageBeddings("xrayexam102.jpg")
insert into health values ('embeddings', '2016-03-05')
embeddings = getImageBeddings("xrayexam103.jpg")
insert into health values ('embeddings', '2016-10-18')
embeddings = getImageBeddings("xrayexam104.jpg")
insert into health values ('embeddings', '2017-02-06')
embeddings = getImageBeddings("xrayexam105.jpg")
insert into health values ('embeddings', '2018-05-20')
embeddings = getImageBeddings("sample.jpg")
select similarity(v, 'embeddings', 'type=euclidean_fraction_float,topk=3')
from health
where year between '2000-01-01' and '2020-12-31'
You can use this technique to detect anomalies or unusual patterns in time-series data by searching for data points that deviate significantly from the norm within a specific time window. It can be applied to recommendation of products, content, or services to users based on their past preferences within a certain time frame. In finance, this approach is useful for identifying similar market trends or trading patterns during specific time intervals. In medical applications, you can search for similar patient health records within a specific timeframe to identify cases with similar symptoms or medical histories.
|
